Research On Data Flow Identification And Classification Technology Based On SSL Protocol

With the rapid development of computer network technology and continuous growth of Internet scale, various network services are springing up with the security issues becoming more and more obvious.Network administrators need to know about the user’s online activities so as to provide better bandwidth request and service, network service providers need to understand the development of web services so as to provide better network development strategies, the scholars need to study network resource planning and network transmission standards, all these are inseparable with the technology of network data flow identification and classification.The SSL protocol provides encryption support for the network data transmission, but since the current data flow recognition and classification algorithms are based on the port or load content, which have brought about challenges for SSL data flow identification and network service classification.The thesis mainly makes the research from the following aspects:First, the thesis makes an in-depth research on traditional data flow classification algorithm and encrypted data flow classification algorithm, and puts forward machine learning data flow classification method based on the characteristics of SSL protocol.Then, by studying the principle SSL protocol and network interaction, the thesis presents extract method of universal SSL protocol, makes computational model of SSL protocol recognition theoretically and verifies the correctness of the method through experiments.Then, by analyzing the commonly used machine learning classification algorithms and comparing the classification accuracy and speed of the algorithms, the thesis chooses the excellent decision tree algorithm as the classification algorithm.In the end, the thesis designs and realizes a data flow recognition and classification system for the SSL protocol on the basis of the existing platform, and achieves classification accuracy of over90%; by the final optimizing of the system, the system achieves satisfactory result in the SSL data flow classification in high-speed network environment.